DeepSeek V4.1 Flash 1017 vs Step 5 Preview
DeepSeek V4.1 Flash 1017 is 2.7× cheaper for typical traffic ; Step 5 Preview scores higher on the Artificial Analysis intelligence index (43.7 vs 39.5).
Prices checked Oct 21, 2026
| DeepSeek V4.1 Flash 1017 | Step 5 Preview | |
|---|---|---|
| Developer | DeepSeek | Stepfun |
| Input / 1M tokens | $0.30 | $1.00 |
| Output / 1M tokens | $1.20 | $2.70 |
| Cached input / 1M | $0.0060 | $0.10 |
| Context window | 1.05M | 1.05M |
| Max output | 393K | 64K |
| Intelligence index | 39.5 | 43.7 |
| Image input | Yes | Yes |
| Tool calling | Yes | Yes |
| Reasoning mode | Yes | Yes |
| Open weights | Yes | Yes |
Monthly cost for common workloads
| Workload | DeepSeek V4.1 Flash 1017 | Step 5 Preview |
|---|---|---|
| Customer chatbot 1,000 conversations a day, 8 turns each | $145 | $392 |
| Document Q&A (RAG) 500 questions a day over retrieved documents | $33.35 | $102 |
| Coding / tool-using agent 200 tasks a day, long tool-heavy contexts | $57.42 | $182 |
| Content generation 30 long articles a day | $3.51 | $8.77 |
| Classification / extraction 100,000 short items a day | $324 | $1,021 |
Estimates include cached-input discounts where a model offers them. Try your own numbers in the cost calculator.
Which one to choose
- Lowest cost: DeepSeek V4.1 Flash 1017 — 2.7× cheaper on a 3:1 input/output mix.